We used R version 3.6.3 (2020-02-29) and the R-packages tidyverse (version 1.3.0), here (version 0.1), brms (version 2.12.0), modelr (version0.1.6), tidybayes (version 2.0.1), and patchwork (version 1.0.0) for data preparation and analysis.

Demographics

We collected data from a total of XX participants. After exclusions based on non-completion or missing data (details to be expanded) we analysed data from 159 participants (age M = 32.65, SD = 8.96, Range = 19, `72). On average participants took 1409.9 seconds to complete the task (SD = 3463.64).

Exclusions: details to be expanded.

Employment

The below graph shows the number of participats in a given employment situation during lockdown.

Lockdown

The below graph shows the number of participants affected by differing lockdown situations.

Living Situation

The below graph shows the number of participants affected by in a given living situation.

Model Fitting

Effect of Hours Played Before and After Lockdown on Mental Health Outcomes

We took the data from the DAS questionnaire ratings before and after lockdown and combined these with hours played before and after lockdown.

Given the data are generated from three Likert-style questionnaire responses per subscale, added together and multiplied by two, responses are thus strictly positive integers. This required fitting the data to cumulative models using a logit link function.

We fitted these models separately for each subscale of the DAS using the brm function in brms, estimating the effect of hours played in video games, time (pre- and post-lockdown), and the interaction between them. The effect of time was sum-coded (i.e. before = 1 and after = -1), such that the fixed effect of time represents the average effect of time across the range of hours played (i.e. a main effect). All models contained random intercepts per participant. Models used a \(Normal(0, 50)\) prior on the intercept, a \(Normal(0, 5)\) prior on the slope terms, and an \(Exponential(1)\) prior on the standard deviation term.

We present the fixed effect parameter estimates below on the log scale followed by the same parameter estimates backtransformed to the natural (i.e. rating) scale. For each model, we present posterior predictions from the model. We evaluate the evidence in support for an effect for each fixed factor using Bayes factors calculated using the Savage Savage-Dickey density ratio with the hypothesis function in brms.

Model Predictions

Below we show spaghetti plots showing the predictions for each model for the effect of hours played within each subscale and time period on DAS outcomes.

Spaghetti Plots showing Draws from the Posterior for the Fixed Effect of Hours Played DAS outcomes Before and After Lockdown (lines). Dots represent observed data.

Spaghetti Plots showing Draws from the Posterior for the Fixed Effect of Hours Played DAS outcomes Before and After Lockdown (lines). Dots represent observed data.

Depression

On the logit scale, model fixed effects parameter estimates are as follows:

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept[1] -4.60 0.58 -5.79 -3.54 1 1303 2397
Intercept[2] -3.06 0.50 -4.08 -2.11 1 1307 2240
Intercept[3] -2.26 0.49 -3.25 -1.33 1 1311 2271
Intercept[4] -1.65 0.48 -2.61 -0.73 1 1356 2374
Intercept[5] -0.90 0.46 -1.84 0.01 1 1385 2322
Intercept[6] 0.02 0.46 -0.91 0.91 1 1437 2456
Intercept[7] 0.50 0.46 -0.40 1.40 1 1495 2442
Intercept[8] 1.06 0.46 0.15 1.98 1 1576 2580
Intercept[9] 1.65 0.47 0.74 2.57 1 1636 2743
Intercept[10] 2.20 0.47 1.26 3.12 1 1669 2920
Intercept[11] 2.60 0.48 1.67 3.55 1 1727 2944
Intercept[12] 3.11 0.50 2.13 4.09 1 1808 2776
Intercept[13] 3.45 0.51 2.48 4.44 1 1821 2993
Intercept[14] 3.82 0.52 2.82 4.84 1 1806 2859
Intercept[15] 4.21 0.53 3.21 5.28 1 1844 2903
Intercept[16] 5.12 0.56 4.05 6.26 1 1975 3195
Intercept[17] 5.58 0.58 4.49 6.74 1 2060 2992
Intercept[18] 6.24 0.62 5.07 7.47 1 2151 2866
Intercept[19] 6.83 0.65 5.62 8.14 1 2266 2879
Intercept[20] 7.68 0.72 6.30 9.15 1 2497 2954
Intercept[21] 8.65 0.82 7.12 10.35 1 2823 2978
time1 -0.76 0.21 -1.17 -0.35 1 3778 3010
total_hours 0.02 0.01 0.00 0.05 1 1921 2987
time1:total_hours 0.01 0.01 0.00 0.03 1 3663 3245

On the natural scale, model fixed effects parameter estimates are as follows:

##              Parameter Estimate Est.Error    Q2.5  Q97.5
## 1       b_Intercept[1]    0.012   0.00673  0.0031  0.028
## 2       b_Intercept[2]    0.050   0.02404  0.0166  0.108
## 3       b_Intercept[3]    0.103   0.04431  0.0374  0.210
## 4       b_Intercept[4]    0.172   0.06577  0.0682  0.325
## 5       b_Intercept[5]    0.297   0.09305  0.1365  0.502
## 6       b_Intercept[6]    0.504   0.10907  0.2875  0.713
## 7       b_Intercept[7]    0.618   0.10352  0.4011  0.803
## 8       b_Intercept[8]    0.733   0.08785  0.5376  0.878
## 9       b_Intercept[9]    0.829   0.06543  0.6766  0.929
## 10     b_Intercept[10]    0.892   0.04602  0.7794  0.958
## 11     b_Intercept[11]    0.925   0.03397  0.8421  0.972
## 12     b_Intercept[12]    0.952   0.02327  0.8942  0.984
## 13     b_Intercept[13]    0.966   0.01741  0.9225  0.988
## 14     b_Intercept[14]    0.976   0.01267  0.9437  0.992
## 15     b_Intercept[15]    0.983   0.00898  0.9611  0.995
## 16     b_Intercept[16]    0.993   0.00402  0.9829  0.998
## 17     b_Intercept[17]    0.996   0.00265  0.9889  0.999
## 18     b_Intercept[18]    0.998   0.00151  0.9937  0.999
## 19     b_Intercept[19]    0.999   0.00091  0.9964  1.000
## 20     b_Intercept[20]    0.999   0.00046  0.9982  1.000
## 21     b_Intercept[21]    1.000   0.00021  0.9992  1.000
## 22             b_time1   -0.759   0.21073 -1.1720 -0.347
## 23       b_total_hours    0.022   0.01323 -0.0042  0.048
## 24 b_time1:total_hours    0.011   0.00759 -0.0039  0.026

We calculated Bayes factors to evaluate the strength of evidence in support of a null effect the fixed effects of total hours played and time (pre- and post-lockdown) and their interaction.

## # A tibble: 3 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <chr>
## 1 (time1) = 0     -0.759    0.211   -1.17     -0.347      0.0205    0.0201 *    
## 2 (total_hours)…   0.0216   0.0132  -0.00421   0.0476    99.9       0.990  <NA> 
## 3 (time1:total_…   0.0110   0.00759 -0.00392   0.0259   233.        0.996  <NA>

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played and the interaction between total hours played (both BF01 >=99). However, we found strong evidence in support of a main effect of time on depression, BF10= 48.75. This supports the notion that post-lockdown measures of depression were higher than pre-lockdown measures.

Anxiety

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept[1] -3.20 0.53 -4.29 -2.25 1 1417 2031
Intercept[2] -1.16 0.46 -2.09 -0.28 1 1675 2235
Intercept[3] 0.00 0.45 -0.90 0.88 1 1850 2491
Intercept[4] 0.70 0.46 -0.19 1.59 1 1924 2521
Intercept[5] 1.71 0.46 0.81 2.60 1 2038 2776
Intercept[6] 2.49 0.47 1.57 3.43 1 1970 2505
Intercept[7] 3.26 0.49 2.32 4.24 1 2032 2624
Intercept[8] 4.04 0.52 3.03 5.09 1 2158 2724
Intercept[9] 4.86 0.56 3.83 5.97 1 2177 2595
Intercept[10] 5.90 0.61 4.74 7.14 1 2269 2736
Intercept[11] 7.04 0.68 5.74 8.40 1 2461 2554
Intercept[12] 7.35 0.70 6.03 8.76 1 2538 2875
Intercept[13] 7.84 0.73 6.45 9.35 1 2521 2946
Intercept[14] 8.65 0.81 7.14 10.25 1 2754 2988
Intercept[15] 9.57 0.92 7.87 11.50 1 3081 2996
Intercept[16] 11.16 1.22 8.99 13.76 1 3740 3189
Intercept[17] 12.81 1.63 9.99 16.44 1 3930 3280
Intercept[18] 24.76 10.28 12.88 52.15 1 5219 3329
Intercept[19] 37.63 14.96 16.79 72.59 1 4500 3157
Intercept[20] 54.08 19.56 23.08 99.39 1 4660 2849
Intercept[21] 80.05 27.47 35.67 139.57 1 4941 3363
time1 -0.18 0.21 -0.59 0.23 1 4144 2859
total_hours 0.03 0.01 0.00 0.05 1 1795 2500
time1:total_hours 0.00 0.01 -0.01 0.02 1 3425 2931

On the natural scale, model fixed effects parameter estimates are as follows:

Parameter Estimate Est.Error Q2.5 Q97.5
b_Intercept[1] 0.04 0.02 0.01 0.10
b_Intercept[2] 0.25 0.08 0.11 0.43
b_Intercept[3] 0.50 0.11 0.29 0.71
b_Intercept[4] 0.66 0.10 0.45 0.83
b_Intercept[5] 0.84 0.06 0.69 0.93
b_Intercept[6] 0.92 0.04 0.83 0.97
b_Intercept[7] 0.96 0.02 0.91 0.99
b_Intercept[8] 0.98 0.01 0.95 0.99
b_Intercept[9] 0.99 0.01 0.98 1.00
b_Intercept[10] 1.00 0.00 0.99 1.00
b_Intercept[11] 1.00 0.00 1.00 1.00
b_Intercept[12] 1.00 0.00 1.00 1.00
b_Intercept[13] 1.00 0.00 1.00 1.00
b_Intercept[14] 1.00 0.00 1.00 1.00
b_Intercept[15] 1.00 0.00 1.00 1.00
b_Intercept[16] 1.00 0.00 1.00 1.00
b_Intercept[17] 1.00 0.00 1.00 1.00
b_Intercept[18] 1.00 0.00 1.00 1.00
b_Intercept[19] 1.00 0.00 1.00 1.00
b_Intercept[20] 1.00 0.00 1.00 1.00
b_Intercept[21] 1.00 0.00 1.00 1.00
b_time1 -0.18 0.21 -0.59 0.23
b_total_hours 0.03 0.01 0.00 0.05
b_time1:total_hours 0.00 0.01 -0.01 0.02

We calculated Bayes factors to evaluate the strength of evidence in support of a null effect the fixed effects of total hours played and time (pre- and post-lockdown) and their interaction.

## # A tibble: 3 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <chr>
## 1 (time1) = 0    -0.183     0.207   -5.92e-1   0.232        16.9     0.944 <NA> 
## 2 (total_hours)…  0.0259    0.0130   7.95e-4   0.0518       50.8     0.981 *    
## 3 (time1:total_…  0.00271   0.00737 -1.17e-2   0.0176      612.      0.998 <NA>

We found overwhelming evidence in support of the null model in comparison to the alternative for all main effects and interactions (all BF01 >= 16). Thus, the data are more likely under the null than the alternative model.

Stress

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept[1] -5.05 0.57 -6.23 -3.98 1 1634 2406
Intercept[2] -4.06 0.53 -5.15 -3.07 1 1568 2390
Intercept[3] -3.27 0.49 -4.26 -2.36 1 1582 2431
Intercept[4] -2.55 0.46 -3.46 -1.68 1 1534 2287
Intercept[5] -1.65 0.44 -2.53 -0.83 1 1556 2414
Intercept[6] -0.89 0.43 -1.77 -0.09 1 1571 2457
Intercept[7] -0.12 0.42 -0.97 0.69 1 1606 2054
Intercept[8] 0.47 0.42 -0.38 1.27 1 1693 2502
Intercept[9] 1.07 0.43 0.21 1.89 1 1750 2200
Intercept[10] 1.84 0.44 0.95 2.69 1 1871 2348
Intercept[11] 2.57 0.45 1.68 3.45 1 2026 2237
Intercept[12] 3.03 0.46 2.12 3.91 1 2117 2651
Intercept[13] 3.56 0.47 2.64 4.48 1 2046 2699
Intercept[14] 4.15 0.50 3.18 5.15 1 2159 3004
Intercept[15] 4.85 0.53 3.83 5.89 1 2296 3287
Intercept[16] 5.33 0.56 4.27 6.47 1 2524 3266
Intercept[17] 5.66 0.58 4.53 6.83 1 2656 2955
Intercept[18] 6.75 0.66 5.51 8.05 1 2936 3366
Intercept[19] 8.13 0.83 6.63 9.81 1 3706 3363
Intercept[20] 9.32 1.06 7.47 11.61 1 4298 3419
Intercept[21] 10.90 1.56 8.29 14.47 1 5746 2827
time1 -0.28 0.20 -0.67 0.10 1 4773 3066
total_hours 0.00 0.01 -0.02 0.02 1 2157 2785
time1:total_hours -0.01 0.01 -0.02 0.01 1 4546 3056

On the natural scale, model fixed effects parameter estimates are as follows:

Parameter Estimate Est.Error Q2.5 Q97.5
b_Intercept[1] 0.01 0.00 0.00 0.02
b_Intercept[2] 0.02 0.01 0.01 0.04
b_Intercept[3] 0.04 0.02 0.01 0.09
b_Intercept[4] 0.08 0.03 0.03 0.16
b_Intercept[5] 0.17 0.06 0.07 0.30
b_Intercept[6] 0.30 0.09 0.15 0.48
b_Intercept[7] 0.47 0.10 0.28 0.66
b_Intercept[8] 0.61 0.10 0.41 0.78
b_Intercept[9] 0.74 0.08 0.55 0.87
b_Intercept[10] 0.85 0.05 0.72 0.94
b_Intercept[11] 0.92 0.03 0.84 0.97
b_Intercept[12] 0.95 0.02 0.89 0.98
b_Intercept[13] 0.97 0.01 0.93 0.99
b_Intercept[14] 0.98 0.01 0.96 0.99
b_Intercept[15] 0.99 0.00 0.98 1.00
b_Intercept[16] 0.99 0.00 0.99 1.00
b_Intercept[17] 1.00 0.00 0.99 1.00
b_Intercept[18] 1.00 0.00 1.00 1.00
b_Intercept[19] 1.00 0.00 1.00 1.00
b_Intercept[20] 1.00 0.00 1.00 1.00
b_Intercept[21] 1.00 0.00 1.00 1.00
b_time1 -0.28 0.20 -0.67 0.10
b_total_hours 0.00 0.01 -0.02 0.02
b_time1:total_hours -0.01 0.01 -0.02 0.01

We calculated Bayes factors to evaluate the strength of evidence in support of a null effect the fixed effects of total hours played and time (pre- and post-lockdown) and their interaction.

## # A tibble: 3 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (time1) = 0    -2.80e-1   0.197    -0.671   0.0968        9.79     0.907 NA   
## 2 (total_hours)… -2.80e-4   0.0125   -0.0244  0.0238      407.       0.998 NA   
## 3 (time1:total_… -7.04e-3   0.00731  -0.0214  0.00708     394.       0.997 NA

Here, we found storng evidence against an effect of time on stress (BF01 = 9.79), and overwhelming evidence against an effect of total hours played and the interaction between time and total hours played on stress (both BF01 > 394).

Change in Hours Played Pre- and Post-Lockdown on Mental Health Outcomes

We next explored the effect of the change in total hours playing games before and after lockdown on the difference in mental health outcomes pre- and post-lockdown. Here, hours played after were subtracted from hours played before, and DAS outcomes after were (separately) subtracted from DAS outcomes before.

Models were again fitted separately for each subscale in brms using the brm function. Here, the data were fitted using a Gaussian model (identity link function), with the fixed effect of difference in hours played. Models used a \(Normal(0, 5)\) prior on the intercept, a \(Normal(0, 1)\) prior on the slope term, and an \(Exponential(1)\) prior on the sigma term.

Again, the presence of an effect of difference in hours played was evaluated using Bayes factors calculated using the Savage-Dickey density ratio. We also present the fixed effects for these models on the natural scale.

We present posterior predictions for the effect of difference in hours played pre- and post-lockdown on differences in outcomes for each subscale. Here, lines represent the posterior median along with 50%, 80%, and 95% credible intervals (shaded).

Posterior Predictions for the Fixed Effect of Difference in Hours Played Pre- and Post-Lockdown on Change in DAS Outcomes. Lines represent the posterior median and 50%, 80%, and 95% Credible Intervals.

Posterior Predictions for the Fixed Effect of Difference in Hours Played Pre- and Post-Lockdown on Change in DAS Outcomes. Lines represent the posterior median and 50%, 80%, and 95% Credible Intervals.

Depression

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -1.7 0.46 -2.61 -0.82 1 3744 2889
hours_diff 0.0 0.03 -0.05 0.05 1 3928 3093
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (hours_diff) …  0.00265    0.0253  -0.0465   0.0519       38.7     0.975 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of difference in hours played pre- and post-lockdown and the difference in depression scores.

Anxiety

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -0.6 0.28 -1.14 -0.03 1 4604 3150
hours_diff 0.0 0.02 -0.03 0.03 1 5147 3074
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (hours_diff) …  0.00398    0.0157  -0.0265   0.0347       60.7     0.984 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of difference in hours played pre- and post-lockdown and the difference in anxiety scores.

Stress

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -1.51 0.39 -2.29 -0.72 1 4005 3014
hours_diff -0.01 0.02 -0.05 0.03 1 3914 3006
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (hours_diff) …  -0.0110    0.0217  -0.0522   0.0327       39.9     0.976 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of difference in hours played pre- and post-lockdown and the difference in stress scores.

Effect of Hours Played During Lockdown on Mental Health Outcomes during Lockdown

Depression

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -2.02 0.69 -3.37 -0.65 1 4251 3033
total_hours_after 0.01 0.02 -0.03 0.05 1 4412 3172
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (total_hours_…  0.00915    0.0187  -0.0274   0.0460       45.2     0.978 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in depression scores.

Anxiety

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -0.64 0.42 -1.48 0.14 1 3868 2737
total_hours_after 0.00 0.01 -0.02 0.02 1 3964 2919
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (total_hours_… -1.89e-4    0.0116  -0.0223   0.0217       84.7     0.988 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in anxiety scores.

Stress

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept -1.07 0.58 -2.21 0.04 1 4589 2986
total_hours_after -0.01 0.02 -0.04 0.02 1 4571 3060
## # A tibble: 1 x 8
##   Hypothesis     Estimate Est.Error CI.Lower CI.Upper Evid.Ratio Post.Prob Star 
##   <chr>             <dbl>     <dbl>    <dbl>    <dbl>      <dbl>     <dbl> <lgl>
## 1 (total_hours_… -0.00986    0.0161  -0.0413   0.0211       49.9     0.980 NA

We found overwhelming evidence in support of the null model in comparison to the alternative for the main effect of total hours played during lockdown and the difference in stress scores.

Model Fitting Diagnostics

Posterior predictive checks were carried out for all fitted models. These show a relatively good fit for each model whereby draws from the posterior are close to the fitted data. This indicates good model fit.

Depression Model 1

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Depression

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Depression

Depression Model 2

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Anxiety Model 1

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Anxiety

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Anxiety

Anxiety Model 2

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Stress Model 1

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Stress

Posterior Predictive Check for the Effect of Hours Played Before and After Lockdown on Mental Health Outcomes for Stress

Stress Model 2

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Posterior Predictive Check for the Effect of Difference in Hours Played Before and After Lockdown on Differences in Mental Health Outcomes for Depression

Loneliness

conent to come.

Gaming Motivations on Loneliness

conent to come.